Bill Brooks Posted January 2, 2017 Author Share Posted January 2, 2017 John: In your example visual recognition software, I think. I see it in my examples as well http://www.alamy.com/stock-photo-kayak-ripples-break-the-surface-of-the-rouge-river-on-a-summer-morning-108969278.html Link to comment Share on other sites More sharing options...
John Mitchell Posted January 2, 2017 Share Posted January 2, 2017 John: In your example visual recognition software, I think. I see it in my examples as well http://www.alamy.com/stock-photo-kayak-ripples-break-the-surface-of-the-rouge-river-on-a-summer-morning-108969278.html That makes sense, so I guess the equation is: same caption + same keywords + shot same day + visual-recognition friendly = BINGO! Link to comment Share on other sites More sharing options...
wiskerke Posted January 3, 2017 Share Posted January 3, 2017 visual-recognition friendly? Have a laugh here. (Still the same as when we first saw this - 06 February 2016 - 08:50 PM.) wim edit: missing link Link to comment Share on other sites More sharing options...
Bill Brooks Posted January 3, 2017 Author Share Posted January 3, 2017 This feature seems to work only for Bill This started me thinking. Why does it work for Bill but not for tarsierspectral. That is because the similars are selected from the first search. When it worked for Bill I was clicking on my 13067 image number. That only searches my portfolio. Then similar would search my portfolio only, and come up only with my images in similar. I will use this feature in the future on my portfolio. In order to review my similar images on any subject to make sure I am not repeating myself So if you search the general collection instead, it comes up with my images and others in the search results. The similars on the buy page also comes up with my similar images and images by others. Here is a buy page starting the search by searching the entire Alamy collection for (Maui beach), and then clicking on one of my images of Po’olenalena Beach. The similar images are about 95% mine but not all mine . My Po’olenalena images have been uploaded over a 12 year time period so uplaosd in the same submission is not as important. I think entry to the similar selection is based on the keywords, particularly the unique beach name keyworded as Po’olenalena. All of the similar images both mine and by others are keyworded either Po’olenalena or Poolenalena. Keywording the unique location name has done the trick. IE complete keywording http://www.alamy.com/stock-photo-poolenalena-beach-on-the-island-of-maui-in-the-state-of-hawaii-usa-29043416.html On another note I should stay away from Po’olenalena Beach in the future. I am obviously oversupplied. So my future solution, to get my images on the similar page, will be to take Alamy’s advice. Shoot useful images, then keyword and caption accurately. I sometimes shoot related images and then worry that they will be broken apart in the search. So this will be another good feature of similar. My solution up to now was to put 2 images into one Alamy image, like this one. This high tide/low tide image has sold several times. Link to comment Share on other sites More sharing options...
John Mitchell Posted January 3, 2017 Share Posted January 3, 2017 visual-recognition friendly? Have a laugh here. (Still the same as when we first saw this - 06 February 2016 - 08:50 PM.) wim edit: missing link That is funny. Well, so much for visual-recognition software. Let's hope that driverless cars work a bit better. Link to comment Share on other sites More sharing options...
John Mitchell Posted January 3, 2017 Share Posted January 3, 2017 This feature seems to work only for Bill This started me thinking. Why does it work for Bill but not for tarsierspectral. That is because the similars are selected from the first search. When it worked for Bill I was clicking on my 13067 image number. That only searches my portfolio. Then similar would search my portfolio only, and come up only with my images in similar. I must have really lucked out then. My Montreal clock tower example (see above) was done using the entire Alamy collection. Link to comment Share on other sites More sharing options...
Bill Brooks Posted January 3, 2017 Author Share Posted January 3, 2017 visual-recognition friendly? Have a laugh here. (Still the same as when we first saw this - 06 February 2016 - 08:50 PM.) wim edit: missing link +1 Wim: Wow. How unsimilar can you get, but the similars are similar. It would appear that “cut out”, and “rear” are common keywords to all of the similar images and the sales image. This visual recognition company lists Alamy as a client http://www.tineye.com/clients Note the feature “Cluster images in your image collection” https://services.tineye.com/DataServices More stuff: https://services.tineye.com/docs TinEye is a Toronto company, must be good. 223 Queen Street East, Toronto, Ontario, Canada M5A 1S2 Link to comment Share on other sites More sharing options...
wiskerke Posted January 4, 2017 Share Posted January 4, 2017 visual-recognition friendly? Have a laugh here. (Still the same as when we first saw this - 06 February 2016 - 08:50 PM.) wim edit: missing link +1 Wim: Wow. How unsimilar can you get, but the similars are similar. It would appear that “cut out”, and “rear” are common keywords to all of the similar images and the sales image. This visual recognition company lists Alamy as a client http://www.tineye.com/clients Note the feature “Cluster images in your image collection” https://services.tineye.com/DataServices More stuff: https://services.tineye.com/docs TinEye is a Toronto company, must be good. You mean so far it only does images from Canada? ;-) wim Link to comment Share on other sites More sharing options...
Bill Brooks Posted January 4, 2017 Author Share Posted January 4, 2017 All kidding aside. Artificial intelligence applied to choosing pics is in the future. We should prepare ourselves for it. Look at the job it did here: http://petapixel.com/2017/01/03/eyeem-let-ai-pick-top-photos-2016-chose/ Link to comment Share on other sites More sharing options...
John Mitchell Posted January 4, 2017 Share Posted January 4, 2017 All kidding aside. Artificial intelligence applied to choosing pics is in the future. We should prepare ourselves for it. Look at the job it did here: http://petapixel.com/2017/01/03/eyeem-let-ai-pick-top-photos-2016-chose/ Wouldn't we have to see the original collection of images that Al had to work with in order to judge out how good a job he did? I wonder if Al is related to Hal. Hope not. Link to comment Share on other sites More sharing options...
Bill Brooks Posted January 4, 2017 Author Share Posted January 4, 2017 Wouldn't we have to see the original collection of images that Al had to work with in order to judge out how good a job he did? I wonder if Al is related to Hal. Hope not. Yes it is all about the quality of the information and the sample size. Humans sample images all of the time. Some impress humans, some do not. When a human intelligence makes a judgement on an image, it recalls the images in its memory. I like this memory. I do not like that. Artificial intelligence consults it’s image database, and makes a judgement call. Just like a human. For instance if Alamy fed images, and sales info for those images into a database, it would have an AI capable of making a sales judgement on new images. Today AI medical programs have greater diagnostic success rates than doctors. Link to comment Share on other sites More sharing options...
christian58 Posted January 5, 2017 Share Posted January 5, 2017 Actually I am rather pleased I think the search is delivering quite good results, relevant results in many categories. Far better then many competitors anyway. Just waiting for some sales now! haha! Link to comment Share on other sites More sharing options...
Cryptoprocta Posted January 5, 2017 Share Posted January 5, 2017 I've looked at some of my uploads which were uploaded on the same day, all with the same general subject. Most are two angles of the same item (meaning only two pics of each item), but all have most keywords other than the specific in common. None have 'similars', not even the alternative angle, although they are mostly adjacent files in my portfolio. Also some photos of the same minor celebrity uploaded on the same day with most keywords identical. In fact, I've looked back several pages in my own port, clicking on random images where there's an alternate angle and none have similars. (I don't particularly mind, and I didn't check every single pic.) I do see the similars in the OP, for sure. Did people opt in for this, or did I opt out at some past date and have forgotten? Or are they experimenting with a selection of contributors? Link to comment Share on other sites More sharing options...
NYCat Posted January 5, 2017 Share Posted January 5, 2017 I've looked at some of my uploads which were uploaded on the same day, all with the same general subject. Most are two angles of the same item (meaning only two pics of each item), but all have most keywords other than the specific in common. None have 'similars', not even the alternative angle, although they are mostly adjacent files in my portfolio. Also some photos of the same minor celebrity uploaded on the same day with most keywords identical. In fact, I've looked back several pages in my own port, clicking on random images where there's an alternate angle and none have similars. (I don't particularly mind, and I didn't check every single pic.) I do see the similars in the OP, for sure. Did people opt in for this, or did I opt out at some past date and have forgotten? Or are they experimenting with a selection of contributors? I see similars on some but not all of my images. No opting in or out. Paulette Link to comment Share on other sites More sharing options...
Cryptoprocta Posted January 5, 2017 Share Posted January 5, 2017 ... Or are they experimenting with a selection of contributors? I see similars on some but not all of my images. No opting in or out. Paulette Tx for the info. Link to comment Share on other sites More sharing options...
Bill Brooks Posted January 6, 2017 Author Share Posted January 6, 2017 All kidding aside. Artificial intelligence applied to choosing pics is in the future. We should prepare ourselves for it. Look at the job it did here: http://petapixel.com/2017/01/03/eyeem-let-ai-pick-top-photos-2016-chose/ Look Facebook's Artificial Intelligence keywording of images. It is only a start, but it is a start. http://petapixel.com/2017/01/04/chrome-add-shows-facebook-sees-photos/ Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.